Re: [patch] to add device+inode check to ipt_owner.c - HACKED UP

[Gianni Tedesco]

On Wed, 2004-09-08 at 11:39 +0100, Luke Kenneth Casson Leighton wrote:
> ... did i sent a patch?
> 
> did i send a patch??  i don't _think_ so *lol* :)

heh :)

IMO the number of constraints involed here make using this patch fairly
involved (for something security related at least) in that, as you said,
you have to:

 - be careful to use ACCEPT rules only
 - be careful that you do:
    1. remove fw rules
    2. upgrade software
    3. replace rules

plus the fastpath code looks very hairy with at least 3 locks taken and
O(num_tasks * max_fds) unpreemptable in softirq...

There has to be a simpler approach, perhaps passing in a path, looking
up the dentry in current namespace and setting an unused flag in
d_vfs_flags? That way you could just match on skb->sk->sk_socket->file-
>f_dentry.

I don't know enough about VFS to know if that's really possible. I mean
would you need vfsmnt too to make it accurate across namespaces? and if
so, could dcookie infrastructure be used?

-- 
// Gianni Tedesco (gianni at scaramanga dot co dot uk)
lynx --source www.scaramanga.co.uk/scaramanga.asc | gpg --import
8646BE7D: 6D9F 2287 870E A2C9 8F60 3A3C 91B5 7669 8646 BE7D

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/